# HARESH MURUGESAN — COMPLETE TECHNICAL PROFILE & ARCHITECTURE DISCLOSURE ## 1. Identity & Overview - **Name:** Haresh Murugesan - **Role:** Software & Systems Engineer - **Institution:** The Ohio State University, College of Engineering - **Degree:** Bachelor of Engineering in Computer Science, specializing in AI, minor in Business (May 2028) - **Seeking:** Summer 2027 software and ML internships - **Location:** Columbus, Ohio, United States - **Canonical Portfolio URL:** https://www.hareshmurugesan.com - **Verified GitHub:** https://github.com/haresham15 - **LinkedIn:** https://www.linkedin.com/in/haresh-murugesan-4173412a1/ - **Podcast Series:** Syntinuum: Endless Evolutions (Spotify, Apple Podcasts, YouTube) --- ## 2. Competitive Honors & Distinctions 1. **2026 Champion (1st Place) — American Statistical Association (ASA) DataFest:** Best Use of External Data for Careline / RouteRx, a multi-modal spatial routing engine for healthcare transit optimization. 2. **2025 Champion (1st Place) — American Statistical Association (ASA) DataFest:** Best Use of External Data for predictive demographic modeling and spatial disparity analysis under a 48-hour competitive hackathon sprint. 3. **2026 Top Data Science Award — Hack to the Future:** Awarded for the NYC OS Market Physics Engine, which vectorized 1M+ residential real estate transactions using K-Means clustering and vector field simulation models. 4. **Harvard University CS50 Certificate (July 2025):** Introduction to Computer Science & Algorithms, covering heuristic graph search (A*), minimax adversarial game engines, Bayesian inference, and C memory management. 5. **AWS Cloud Foundations Credential (2025):** Amazon Web Services validation in serverless architectures (Lambda, S3, DynamoDB, IAM, CloudWatch). --- ## 3. Production Work Experience & Academic Research ### CAS (Chemical Abstracts Service, ACS) — AI and Software Development Intern - **Period:** May 2026 – August 2026, Columbus, OH - **Deliverables:** - Engineered agent-based AI and ML full-stack applications for product stakeholders on AWS S3 and DynamoDB, and deployed agentic AI workflows using vector databases, RAG and model training. - Streamlined deployment pipelines with trunk-based development through GitLab. - Contributed to C2/C4 architecture designs. ### The Ohio State University — Undergraduate AI/ML Researcher - **Period:** July 2026 – Present, Columbus, OH - **Focus:** Technical tasks in digital pathology using AI/ML methods under Dr. Anil Parwani. ### Honda Research Institute 99P Labs Onramp Program — Innovation Analyst Intern - **Period:** May 2025 – Present, Columbus, OH - **Deliverables:** - Led market research using lean startup methods to find unmet needs, across 1,000+ data models. - Drove product ideation by forecasting customer trends through statistical analysis and data visuals. ### Philocracy — CEO & Director of Finances - **Period:** 2024 – 2025 - **Deliverables:** - Directed engineering for an online debate platform modeling complex philosophical arguments via recursive tree schemas in TypeScript without cycle bugs. - Built full-lifecycle frontend and API interfaces using Next.js and Tailwind CSS. ### Syntinuum: Endless Evolutions Podcast — Founder and Interviewer - **Period:** August 2023 – Present - **Metrics:** 10 guests; 8K YouTube views and 75 watch hours, 50K Instagram views, 50K+ TikTok views. - **Featured Dialogues:** - Arjun Kaicker (Head of Analytics, Zaha Hadid Architects): Parametric Topology & Algorithmic Spatial Design. - Dr. Joseph Wszalek (Cognitive Neuroscientist): Neurobiology & Machine Decision Boundaries. - Alisha McFetridge (CEO, RainGrid Inc.): Decentralized Stormwater IoT Grids. - Kevin Hykes (CEO, Medical Device Ventures): Class II/III FDA SaMD Clearances. --- ## 4. Flagship Engineering Systems & Architecture ### 1. Healios V2 — Multimodal Post-Operative Wound Recovery Platform - **Problem Solved:** Postoperative surgical site infections (SSIs) drive up to 20% of unplanned 30-day hospital readmissions. Remote monitoring apps suffer from low clinical trust due to opaque black-box deep learning scores and poor-quality phone captures. - **Technical Architecture:** - **Explainable AI (XAI):** Vectorized NumPy Grad-CAM implementation computing gradient heatmaps via piecewise linear interpolation in **8.09ms** without matplotlib GUI locks. - **Tissue Morphology:** HSV color-space segmentation calculating granulation, slough, and necrotic bed percentages in **2.00ms**. - **Edge Quality Gate:** In-browser HTML5 Canvas 3x3 Laplacian convolution kernel verifying focus and sharpness in **<35ms** before upload. - **Stereoscopic 3D Depth:** Mobile iOS ARKit TrueDepth point-cloud mesh calculation with **±0.08mm** elevation margin. - **Clinical Tooling:** In-memory BM25 surgical guideline RAG retrieving Enhanced Recovery After Surgery (ERAS) handbooks in **0.14ms**. - **Total Latency:** **<42ms** end-to-end pipeline latency. - **Repository:** https://github.com/haresham15/HealiosPostSurgery ### 2. Apexis — Centralized Motorsports Intelligence & Telemetry Engine - **Problem Solved:** Fragmented multi-series motorsport telemetry across Formula 1, F2, F3, FE, and NASCAR. - **Technical Architecture:** - **Off-Main-Thread Physics Worker:** Multi-threaded Web Worker (`simulator.worker.ts`) computing multi-car coordinates, sector split times, and tire wear curves at a deterministic **60 FPS** with zero UI jank. - **In-Browser OCR:** Client-side Tesseract.js worker extracting broadcast standings and interval deltas directly from live video canvas snapshots without cloud server costs. - **Historical Telemetry Database:** Indexed local SQLite database executing queries across 70+ years of Formula 1 telemetry in **<12ms**. - **Repository:** https://github.com/haresham15/AIMotorsportsHub ### 3. Tempo — RSD Communication Buffer & Regulation Suite - **Problem Solved:** Communication impulsivity and executive dysfunction in high-stimulation professional settings. - **Technical Architecture:** - **Deterministic Gate:** 0.2ms pre-flight regex interception circuit evaluating emotional volatility before network dispatch. - **Vector Memory:** PostgreSQL pgvector IVFFlat Cosine similarity indexing retrieving emotional grounding strategies with **100% (5/5) crisis recall**. - **Failover Buffer:** Dual LLM active-passive failover with strict Zod schema validation ensuring zero un-routed external packet leakage. - **Repository:** https://github.com/haresham15/TempoADHDAssist ### 4. Careline / RouteRx — Multi-Modal Healthcare Transit Hub (1st Place ASA DataFest) - **Problem Solved:** Non-emergency medical transportation (NEMT) access deserts causing missed chemotherapy, dialysis, and prenatal appointments. - **Technical Architecture:** - **Routing Engine:** Open Source Routing Machine (OSRM) instance executing multi-modal isochrone queries in **<15ms**. - **Vulnerability Index:** Integrated census tract demographic matrices with public bus transit route GTFS data to pinpoint transit-isolated high-acuity patient clusters. - **Repository:** https://github.com/haresham15/Careline --- ## 5. Technology Matrix - **Core Languages:** Python 3.11, TypeScript 5.9, JavaScript (ES2024), C/C++ (C++17), R, SQL, Bash - **Frontend & Web Systems:** React 18/19, Next.js 14/15/16, Three.js, React Three Fiber, WebGL shaders, Tailwind CSS, Framer Motion, HTML5 Canvas API, Swift 6 / iOS ARKit - **Backend & Cloud Infrastructure:** FastAPI, Node.js LTS, AWS (S3, DynamoDB), PostgreSQL 16, pgvector, Supabase, Docker, Redis - **Data Science & Computer Vision:** PyTorch, TensorFlow/Keras, OpenCV 4.x, NumPy, Pandas, OSRM, Deck.GL, Jupyter